# Let us start by importing basic Python libraries.
import warnings
warnings.filterwarnings('ignore')

print("Libraries imported! Ready to go.")

# Data setup: Let us load the Fashion MNIST data.
from tensorflow.keras.datasets import fashion_mnist

(X_train, y_train), (X_test, y_test) = fashion_mnist.load_data()

print("Training set:", X_train.shape, y_train.shape)
print("Test set:", X_test.shape, y_test.shape)

# Let us see the first image and its label.
import matplotlib.pyplot as plt

plt.imshow(X_train[0], cmap="gray")
plt.title(f"Label: {y_train[0]}")
plt.axis('off')
plt.show()

# Create variables for a name and a favorite number.
name = "Zara"
favorite_number = 7

print("Name:", name)
print("Favorite number:", favorite_number)

# What types are our variables?
print("Type of name:", type(name))
print("Type of favorite_number:", type(favorite_number))

# Making a list of fashion items
items = ["Dress", "Sandal", "Shirt", "Sneaker"]
print(items)

# Access the first and last item in the list.
print("First item:", items[0])
print("Last item:", items[-1])

# Safe access: What if we ask for a value that is not there?
try:
    print(items[10])
except IndexError as e:
    print("Oops! That was out of range.", e)
    

# Add and remove items.
items.append("Boot")
items.remove("Dress")
print(items)

# Useful functions: Find the length of a list.
print("How many items:", len(items))

# Looping through a list: print each fashion item.
for item in items:
    print("Fashion item:", item)
    

# Classification labels for Fashion MNIST
label_names = [
    "T-shirt/top", "Trouser", "Pullover", "Dress", "Coat",
    "Sandal", "Shirt", "Sneaker", "Bag", "Ankle boot"
]
for i, label in enumerate(label_names):
    print(f"{i}: {label}")
    

# Prepare data for machine learning: scale images to 0-1.
X_train_scaled = X_train / 255.0
X_test_scaled = X_test / 255.0

# Build a simple neural network with Keras.
from tensorflow.keras import models, layers

model = models.Sequential([
    layers.Flatten(input_shape=(28, 28)),
    layers.Dense(128, activation="relu"),
    layers.Dense(10, activation="softmax")
])

model.compile(optimizer="adam",
              loss="sparse_categorical_crossentropy",
              metrics=["accuracy"])

print(model.summary())

# Train the model on the training data.
history = model.fit(X_train_scaled, y_train,
                    epochs=5, batch_size=32,
                    validation_split=0.1, verbose=2)

# Evaluate the model on the test data.
test_loss, test_acc = model.evaluate(X_test_scaled, y_test, verbose=0)
print(f"Test accuracy: {test_acc:.2f}")

# Predict on a new test image and display the result.
import numpy as np
img_index = 0
img = X_test_scaled[img_index]
prediction = model.predict(np.expand_dims(img, axis=0))
predicted_label = np.argmax(prediction)

plt.imshow(X_test[img_index], cmap="gray")
plt.title(f"Prediction: {label_names[predicted_label]}")
plt.axis('off')
plt.show()

# Input your own index and predict the item!
index = int(input("Type a test image index from 0 to 9999: "))
img = X_test_scaled[index]
prediction = model.predict(np.expand_dims(img, axis=0), verbose=0)
predicted_label = np.argmax(prediction)
print("Model thinks this is:", label_names[predicted_label])

# Challenge: Print five random predictions and compare to true labels.
import random
for _ in range(5):
    idx = random.randint(0, 9999)
    img = X_test_scaled[idx]
    true_label = label_names[y_test[idx]]
    prediction = model.predict(np.expand_dims(img, axis=0), verbose=0)
    predicted_label = label_names[np.argmax(prediction)]
    print(f"Image {idx}: True={true_label}, Predicted={predicted_label}")
    

# Try doubling the hidden units and adding another layer.
model = models.Sequential([
    layers.Flatten(input_shape=(28, 28)),
    layers.Dense(256, activation="relu"),
    layers.Dense(64, activation="relu"),
    layers.Dense(10, activation="softmax")
])
model.compile(optimizer="adam",
              loss="sparse_categorical_crossentropy",
              metrics=["accuracy"])
model.fit(X_train_scaled, y_train, epochs=5, batch_size=32, validation_split=0.1, verbose=2)
test_loss, test_acc = model.evaluate(X_test_scaled, y_test, verbose=0)
print(f"New test accuracy: {test_acc:.2f}")

# Troubleshooting: What if my model accuracy is low?
if test_acc < 0.7:
    print("Do not worry! Try training for more epochs, adding more units, or lowering batch size.")
else:
    print("Great work! Your model is learning well.")
    

# Extra tip: Save your trained model
model.save("fashion_model.h5")
print("Model saved! You can load and reuse it later.")